The Agent Skills Directory

[DATA_EXFILTRATION]: The skill performs broad searches for highly sensitive documents containing payroll, salary, compensation, and benefits information. While no network exfiltration is explicitly coded, accessing these files represents a significant data exposure risk.
[COMMAND_EXECUTION]: The skill uses bash commands (find, grep, xargs) to scan the filesystem for sensitive financial and personnel records.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests data from external sources (CSV and JSON files) that may be controlled by untrusted parties. These files are processed without explicit sanitization or boundary markers to prevent embedded instructions from influencing agent behavior.
Ingestion points: Reads files found via find command in Step 0.
Boundary markers: None used when processing file content.
Capability inventory: Uses Bash, Read, and Grep which can be leveraged if an injection occurs.
Sanitization: None detected.

mint-budget